Short answer: a defect-reduction number is only worth something when it comes with a baseline, a failure definition, a production volume and a record of what else changed. This page is the method for producing that evidence on your own line — and for testing any supplier's claim, including ours.
A note on this page. It previously carried a headline claiming a specific customer result. We could not support that figure with a named baseline, a documented trial or customer authorisation, so we replaced it with the method instead of restating it. When a customer gives us written permission to publish their data, we will publish the case with its sources — and until then we would rather be checkable than impressive.
1. Define the defect before you change anything
"Defects" is not a measurement. Harness tape problems need separating, because they have different causes and different fixes:
| Failure mode | Typical cause to investigate | Where it is detected |
|---|---|---|
| Flagging at the tape end | Adhesive relaxing after heat, wrong end-fixing method, overlap too short | After thermal cycle, transport, or at the vehicle plant |
| Edge lift at a branch | Conformability, tension, bundle diameter, surface contamination | Final inspection |
| Abrasion through the wrap | Abrasion class below the drawing requirement, routing against an edge | Durability test or field |
| Noise / rattle complaint | No damping grade where the loom touches trim | Vehicle assembly or customer |
| Residue on rework | Adhesive system, time at temperature, removal method | Repair operations |
Pick the one you intend to reduce. A trial aimed at "defects" generally proves nothing; a trial aimed at flagging in the bulkhead zone can.
2. Establish a baseline from your own records
Before any sample arrives, pull from your own data:
- the period you are using as the baseline, with dates;
- the volume produced in that period;
- the count of the specific failure mode, not of all rejects;
- the inspection criterion used to call it a failure, in writing;
- the tape lot numbers in use during that period.
If the baseline rate is so low that it appears a handful of times a month, say so now — it changes how long the trial has to run, and it is better to discover that before the trial than after.
3. Change one thing
A controlled trial changes the tape and nothing else: same harness variant, same fixture, same work instruction, same overlap, same operators where the roster allows, same inspection criteria, same cable lot if possible. Record the lot number of both the incumbent and the candidate tape.
If the work instruction has to change — a different overlap, a different end-fixing — that is a second variable, and the result can no longer be attributed to the tape alone. Run it as a second trial.
4. Run it long enough to count
Decide the run length from the baseline rate, before starting. If the failure appears in roughly one unit in two hundred, a fifty-unit trial cannot distinguish an improvement from an ordinary quiet week. Where the rate is low, options are:
- run longer, across more than one shift and more than one operator;
- include an accelerated condition — the thermal cycle or transport simulation where the failure actually appears;
- measure a precursor instead: edge-lift length at inspection, rather than the final reject.
5. Record what an auditor would ask for
- Written trial plan, dated, naming the failure mode and acceptance criterion.
- Baseline extract from the production records, with period and volume.
- Lot numbers and data sheets for both tapes.
- Inspection results classified by failure mode, not aggregated.
- Photographs of representative failures, before and after.
- Change record confirming nothing else moved during the trial.
- Decision and sign-off, with who approved it.
That package does three jobs: it settles the internal argument, it survives a customer audit, and it gives you a defensible baseline for the next change.
6. How to test a supplier's claim
When a supplier quotes a percentage, ask four questions:
- Over what baseline period and volume?
- Which failure mode, defined how?
- What else changed at the same time?
- Is the data the customer's, and is it authorised for publication?
Those four questions are also the reason this page no longer carries a number of its own.
FAQ
Why distrust a defect-reduction percentage?
Without baseline, failure definition, volume and a record of other changes, it cannot be checked — ours included.
What makes a trial controlled?
One variable changes. Same harness, fixture, instruction, criteria; both lot numbers recorded.
How long should it run?
Long enough for the failure to occur often enough to count — decided from the baseline rate before starting.
What evidence survives an audit?
Dated baseline, written plan, both lot numbers, data classified by failure mode, photographs, and a change record.
Related: where harness tape costs you money, qualifying a tape supplier, harness tape types and how to choose.